Case Study: Mattermark achieves scalable, cost-efficient AWS resource utilization with Mesosphere

How Mattermark uses Mesos and Marathon to make the most of AWS

Mattermark, a startup that provides data on private companies, was running on just a handful of AWS instances but struggled with operational chaos: hundreds of daily data-mining, machine-learning and indexing jobs were launched haphazardly, undocumented, and it was often unclear which EC2 instance was running what. With plans to scale its database and add new data types, the company needed a more reliable, scalable way to manage resources and schedule work.

Mattermark re-architected its infrastructure using Apache Mesos with Marathon and Chronos, containerized its apps with Docker, and gained an abstraction layer, fine-grained scheduling, and resource isolation. The result: automated, reliable job scheduling, smarter packing of diverse workloads onto fewer instances (targeting 80–90% utilization), lower costs, and an easier path to add technologies like Kafka and Spark and scale the business — with plans to simplify operations further using Mesosphere DCOS.

Mattermark

Samiur Rahman

Machine Learning Engineer